| name | weekly-review |
| description | Weekly synthesis of your brain thoughts, workflow insights, and goals. Surfaces gaps, open loops, and recommendations. Every claim is cited back to its source. |
Weekly Review
A weekly synthesis that cross-references your brain thoughts, workflow insights, and goals to surface what you'd miss looking at any one source alone.
Every claim in the output must be grounded in a source — cite thoughts as thought:<id>, insights as insight:<id>, lists as list:<id>.
Workflow
Step 1: Establish the Week's Time Window
Compute the start of the week as an epoch-ms timestamp.
- "This week" = from 7 days ago at 00:00 local time to now.
- Convert to epoch ms:
Date.now() - 7 * 24 * 60 * 60 * 1000 (rounded down to the start of the day).
Call this weekStartMs.
Step 2: Pull the Week's Timeline
Call mcp__ai-brain__timeline_thoughts with:
aroundMs: weekStartMsbefore: 0after: 50
This returns up to 50 compact index rows in chronological order for thoughts captured since the start of the week.
If the array is empty, tell the user: "Your brain has no thoughts captured this week. Try capturing some thoughts, or run /brain-init if your brain is empty." Then stop.
Step 3: Pull Insights and Goals
Workflow insights:
Call mcp__ai-brain__get_insights with status: "new", then again with status: "noted". If the tool is unavailable (server doesn't expose insights for this user), note this and skip insight-dependent sections.
Goals and priorities:
Call mcp__ai-brain__get_lists with pinned: true to get the user's stated goals and priorities.
Open items:
Call mcp__ai-brain__get_open_items to get unfinished tracked items across all lists.
Step 4: Hydrate Noteworthy Thoughts
From the week's timeline (Step 2), select up to 10 thoughts that look most substantive (by summary + snippet + topics) — the ones you'll want to cite in the synthesis.
Call mcp__ai-brain__get_thoughts with ids: [<up to 10 ids>]. Only these hydrated results can be quoted or paraphrased in the synthesis. The rest of the timeline is still referenceable by thought:<id> + summary.
Step 5: Produce the Review
Generate a report with these 5 sections. Each section should be concise — the entire review should be scannable in 2 minutes. Every factual claim must cite its source.
This Week in 30 Seconds
2-3 sentence headline summary. What was the dominant theme? What stands out?
Cite 2-3 anchor thoughts: thought:<id>.
Attention vs. Intention
Compare workflow insights (what you actually did) against pinned goals (what you intended to do).
Flag:
- Goals with no corresponding session activity — "You said [goal] is a priority but had no sessions related to it" — cite the goal as
list:<id>.
- Heavy activity on topics not in your goals — "[Topic] consumed [X]% of sessions but isn't in your goals" — cite
insight:<id>.
- Momentum shifts — "[Topic] went from [X]% to [Y]% of sessions week over week" — cite
insight:<id>.
If no workflow insights are available, display instead:
"Install the radar plugin (/plugin install radar@flippyhead/radar) for time allocation analysis."
Knowledge Captured
Review the themes from saved thoughts this week. Cite each theme to an anchor thought.
Highlight:
- Topics with workflow insights but no saved thoughts — "You worked on [topic] but didn't save any knowledge about it — is there something worth persisting?" — cite
insight:<id>.
- Repeated thought topics — building momentum on a theme — cite 2-3
thought:<id> examples.
- Cross-domain connections — thoughts from different contexts that might be related — cite the connected thoughts.
Open Loops
Aggregate unfinished threads from all sources:
- Open items from pinned lists — cite each
list:<id> with item counts.
- Workflow insights still marked "new" — cite each
insight:<id>.
- Decisions mentioned in thoughts that lack clear resolution — cite the
thought:<id> where the decision was raised.
Next Week
2-3 specific, actionable recommendations based on the above. Be forward-looking, not retrospective. Reference specific projects, people, or decisions — with citations — when possible.
Step 6: Offer to Save
After presenting the review, ask:
"Want me to save a summary of this review to your brain? This helps track trends across weeks."
If yes, save a condensed version via mcp__ai-brain__capture_thought with format:
"Weekly review (week of [ISO date]): [2-3 sentence summary of key themes, attention vs. intention highlights, and top recommendation]. Grounded in: thought:, thought:, insight:."
Return the new thought:<id> to the user so they can find this review later via /brain-thread or /brain-context.
